search

wangqixun / mfpsg

mask2former psg
Apache License 2.0
22 stars 1 forks source link

Some problems about ablation study and loss mode #2

Open DrFV opened 11 months ago

DrFV commented 11 months ago

Thank you for your excellent work! I am currently focusing on the PSG. I noticed in your paper, '1st Place Solution for PSG competition with ECCV'22 SenseHuman Workshop,' that you emphasize the importance of focal loss and self-distilled soft labels. However, I couldn't find the results of the ablation study. Could you please provide more detailed information? Additionally, I have observed that when applying both focal loss and self-distilled soft labels, the batch size must be set to 1. Is there a method to increase the batch size while using both of these tricks?

Vincent-luo commented 9 months ago

@DrFV I spent a long time looking for the code for self-distillation and still couldn't find it. Can you tell me where I can find the related code? Thanks!